[Wien] GGA and lanthanides

[Peter Blaha]

I would like to weaken Stefaan's statement.

The importance of SO and LDA+U depends a lot on the properties you are
looking for. In particular in Steffan's field of hyperfine interactions
SO is probably mandatory, while for other properties it might be a fairly
small effect.
My suggestion is: 

start out with a regular GGA calculation. If the results look good - fine.

Since the 4f electrons in GGA are always "pinned" at EF and never really
"localize", LDA+U is probably quite important for most lanthanide compounds
(it depends on the character of the compound: metallic or ionic ?; type of
element (half-filled f-shell ?).

And of course, spin-orbit may also influence the results, but in general
it will be a smaller effect (the SO splitting of the 4f shell is relatively
small) than LDA+U.

> 
> > Hello. I'd like to know if it is alright to use the GGA method to study
> the structural and
> > magnetic phase stability of lanthanide metallic compounds? I have already
> applied it ( not
> > including SO ) to many cases and the results are in very very good
> agreement with the
> > existing experimental data.
> 
> I would doubt... SO for lanthanides is mandatory, and if there are localized
> magnetic moments in your compounds, then also LDA+U must be used. Try to
> compare as many properties as you can with experiment : sometimes a good
> agreement for one or two properties is accidental. Check the UG as well for
> "Relativistic Local Orbitals" (RLO), which may be appropriate for your
> compounds.


                                      P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------